Then what does gzip do when it reads on stdin? (maybe that fixes your rsync problem) On Mon, Apr 28, 2003 at 05:32:46PM +0200, Axel Thimm alleged: > On Mon, Apr 28, 2003 at 11:08:00AM -0400, seth vidal wrote: > > On Mon, 2003-04-28 at 03:49, Axel Thimm wrote: > > > I observed that recreating the headers of the same set of rpms is creating > > > *.hdr files that are different in content. Does some date enter the *.hdr > > > files? > > > > > > Would it be possible to have the same *.hdr files produced (maybe even > > > timestamp them with the time of the rpm itself)? The reason I am asking is > > > that it may save some rsync time for repository mirroring (e.g. when I build a > > > new rpm and recreate the *.hdr files I have to rsync all *.hdr to the master > > > repo) and also improve caching (yummers with a cache don't need to reget all > > > *.hdr files because I added only one new). > > > > so I found out a bit more on this after playing a bit. > > > > it's not the rpm functions that are causing the variation - its gzip. > > > > gzip even acts like this from the command line. > > > > I'll see what else I can find out - but it looks like this may be > > unavoidable. > > I think gzip includes the modification time of the included file. If the > modification time of the created hdr files were set to that of the rpm itself, > gzip would always yield the same result, and it would also make > rsyncing/caching the hdr files even simpler! :) > -- > Axel.Thimm@xxxxxxxxxxxxxxxxxxx
